B0919
Title: Sure Independence Screening (SIS) for multiple functional regression model
Authors: Yuan Yuan - Auburn University (United States) [presenting]
Nedret Billor - Auburn University (United States)
Abstract: Due to rapid advancements in computer technology, high-dimensional, big and complex data, such as functional data where observations are considered as curves, have emerged from applications of biomedicine, chemometrics, engineering, and social sciences. Since functional data are inherently infinite-dimensional, variable selection problem in multiple functional regression model is, therefore, challenging and difficult. A novel approach is offered for functional feature selection under high-dimensional context based on Sure Independence Screening with functional predictors and scaler responses. High dimensionality means that $p = O(\text{exp}(n^{r}))$, where $p$ is the number of functional predictors, and $n$ is the sample size. With simulation studies and real data application, the current method detects true feature set among thousands of functional features and show potential in high-dimensional functional classification as well.
